- 學位論文
從使用者音樂聆賞歷程探討線上音樂分類架構
Exploring User-oriented Taxonomy of Online Music Databases from Music Listening Processes
摘要
使用者導向的線上音樂分類架構，可創造良好的瀏覽經驗，並觸發使用者的音樂需求；所謂使用者導向，應整合資訊行為整體脈絡，及資訊系統與特定資訊行為間的關係，以確保資訊系統設計與評估的完整性及應用性。本研究從使用者音樂聆賞整體歷程出發，瞭解使用者分類相關經驗，並提出線上音樂分類架構具體改善建議。在研究設計上，本研究分別從系統面及使用者面抽取音樂分類的要素、概念及行動，分析比對兩方面所得資料，以瞭解兩者間之差異及關聯。 系統面，針對立意抽樣及滾雪球抽樣取得的31個國內外線上音樂網站 / 資料庫之音樂分類架構進行內容分析。分析結果發現，線上音樂分類架構主要的分類層面為音樂類型、歌手 / 藝人、語言 / 地區及情境。在俗民分類的部分，分析Last.FM的熱門標籤，發現「音樂類型」仍是社群最常使用的標籤類型，其次為「語言 / 國家」、「情緒感受」及「個人」。此外，線上音樂分類架構具有「分類依據缺乏規則」、「類目名稱同義異字」、「分類隸屬不一致」、「以行銷為目的之類目安排」以及「英文音樂網站 / 資料庫分類依據多元」等現象。 使用者面，本研究邀請31位音樂愛好者參與研究，蒐集的資料包括：為期10∼30日份的音樂日誌（31位）、深度訪談記錄（25位）以及個人數位音樂蒐藏檔案列表（18位）。從資料中，發現音樂愛好者音樂聆賞歷程是由音樂聆賞行為、音樂尋求行為以及個人音樂組織行為所構成。而音樂聆賞歷程中，「分類」所扮演的角色，包括個人音樂蒐藏管理方式、音樂尋求輔助工具、音樂聆賞需求引動媒介以及聆賞者心理活動狀態表現。此外，本研究從31位音樂愛好者音樂日誌以及2,397篇網路音樂社群音樂尋求問答資料中，分析使用者描述音樂的構面，找出音樂潛在的分類方式。綜合上述使用者面之研究發現，歸結出使用者音樂分類「音樂固有屬性」及「音樂互動特性」兩大分類構面中各分類層面。 最後，分析比較音樂分類層面及類別項目，發現使用者分類與系統分類在「分類層面間連結對應關係」及「分類層面之應用功能」兩方面具有異同及關聯之處。 根據研究發現及研究結果，提出研究建議。在線上音樂資料庫之分類，建議系統應提供「完整描述型分類」以輔助使用者尋求特定音樂；此外「多元主題型分類」可輔助使用者尋求非特定音樂並引動潛在需求。針對個人數位音樂蒐藏分類，提出「將辨識音樂基本資訊標記於數位音樂檔案中」、「善用音樂分類層面的複分特性，減少垂直的階層數」、「彈性運用『暫存 / 其他』分類類別」、「界定『語言 / 地區 / 國家』分類層面之定義」以及「選擇功能齊全的音樂組織軟體」等建議。本研究屬於探索性研究，以探索過程及研究發現中的不足與限制為基礎，針對後續研究，提出「跨學門研究音樂分類」以及「結合多元研究方法探究音樂分類相關概念」兩項建議。
並列摘要
A user-oriented taxonomy for online music databases is able to improve the browsing experience and trigger user's needs. In order to ensure the integrity and applicability in information systems design and evaluation, a user-oriented approach is supposed to integrate both the overall context of information behavior and specific behavior in relation to the information systems. Based on the overall music listening process, this study tries to explore classification-related user experiences and provide constructive suggestions to improve online music taxonomies. In the study design, this study extracts and analyzes elements, concepts and operations of music categorization from the music information retrieval (MIR) systems and users in order to realize the differences and relations between systems and users. To discover the current developments in taxonomies for online music databases, this study analyzes 31 domestic and international online music databases collected by purposive sampling and snowball sampling techniques. The results show that music genre, artist, language/region, and context are the most widely used facets of the online music taxonomies. An analysis of the “Top Tags” list in Last.FM system shows that the “music genre” is the most commonly used type of tags which are produced by online community members, followed by “language/country”, “mood” and “personal” types of tags. In addition, this study explores the state of current developments in online music taxonomies, including “no exact rules in classification schemes”, “synonymous categories named differently from others”, “inconsistencies in the classification hierarchy”, “category arrangements based on marketing purposes” and “classification schemes of overseas music databases use a variety of facets to classify music collections.” To investigate users’ overall context of music listening, 31 music lovers were invited to participate in this study and provided music dairies which recorded users’ everyday musical experience for 10-30 days, in depth interview records and the list of personal digital music collections. Results indicated that the music listening process is consisted of three dimensions: music listening behavior, music seeking behavior and the organization of personal music collection. The roles of classification in the whole music listening process can be identified as methods of personal music management, music seeking aids, triggers of music listening needs and expressions of listener’s psychological activity. Besides, this study analyzes descriptors of music lovers’ music dairies and 2,397 posts of online music community in order to discover potential facets of music classification schemes. To sum up, there are two main dimensions in music classification facets of data collected: “the inherent properties of music” and “interactive attributes of music.” At the end, the comparative analysis of music classification facets and categories found that differences, similarities and relations between users’ classifications and taxonomy lays in “the linkage relations of classification facets” and “application functions of classification facets.” Suggestions for taxonomies of online music databases and classifications of users’ personal music collections are proposed. Online music databases are supposed to provide “complete bibliographic descriptors of music” that can help users find specific music items. In addition, “multiple theme-based classifications” can also help users find non-specific music items and trigger potential music needs. Suggestions for organizing personal digital music collections include “labeling basic bibliographic data in the digital music files”, “making the best use of music classification facets’ subdivision properties to reduce the hierarchy of classification structures”, “flexible use of the temporary/other category”, “defining the meanings of language/region/country classification facets”, and “choosing a full-featured music organization software.” Based on the limitations of this study, two suggestions were provided for further research – “research music taxonomies in a cross-discipline view” and “combine multi-approach to investigate concepts related to music classification.”